Papers.

The Reactive Principles

The Rapid Growth of Human-in-the-Loop Machine Learning.

The Implicit and Explicit Regularization Effects of Dropout.

Task-agnostic Indexes for Deep Learning-based Queries over Unstructured Data.

Solving the Bottom Turtle — a SPIFFE Way to Establish Trust in Your Infrastructure via Universal Identity

Simplifying Models with Unlabeled Output Data.

Shape Matters: Understanding the Implicit Bias of the Noise Covariance.

Self-training Avoids Using Spurious Features Under Domain Shift.

Selection via Proxy: Efficient Data Selection for Deep Learning

Robust Multi-Agent Reinforcement Learning with Model Uncertainty

Robust and On-the-fly Dataset Denoising for Image Classification.

Resource Elasticity in Distributed Deep Learning

REPAINT: Knowledge Transfer in Deep Actor-Critic Reinforcement Learning

Ranking and Retrieval under Semantic Relevance

Python 3 for Machine Learning

Proceedings of the 10th International Temporal Web Analytics Workshop (TempWeb 2020), Taipei, Taiwan, April 20

Principled Optimization Of Dynamic Neural Networks

This dissertation generalizes the specialized compilation techniques applied to static dataflow graphs, the predominant programming model of deep learning, to fully dynamic neural networks. Contributions include an intermediate representation for dynamic neural networks, a set of general optimizations as well as specialized dynamic optimizations, and an efficient runtime for dynamic neural networks.

Performance Optimizations and Operator Semantics for Streaming Data Flow Programs

Pancake: Frequency Smoothing for Encrypted Data Stores

Optimal Regularization Can Mitigate Double Descent.

On the Expressivity of Neural Networks for Deep Reinforcement Learning.

Neural Representations for Modeling Variation in English Speech

MultiWOZ 2.1: A Consolidated Multi-Domain Dialogue Dataset with State Corrections and State Tracking Baselines

MOPO: Model-based Offline Policy Optimization.

Model-based Adversarial Meta-Reinforcement Learning.

MLPerf Training Benchmark

Meta-learning Transferable Representations with a Single Target Domain.

Measuring foreign accent strength using an acoustic distance measure

Learning Self-Correctable Policies and Value Functions from Demonstrations with Negative Sampling.

Learning Over-Parametrized Two-Layer ReLU Neural Networks beyond NTK.

Learning Over-Parametrized Two-Layer Neural Networks beyond NTK.

LanguageARC: Developing Language Resources Through Citizen Linguistics

Instruction-level Reliability Improvement for Embedded Systems

Inducing and Exploiting Activation Sparsity for Fast Neural Network Inference

Individual Calibration with Randomized Forecasting.

In-N-Out: Pre-Training and Self-Training using Auxiliary Information for Out-of-Distribution Robustness.

Improved Sample Complexities for Deep Neural Networks and Robust Classification via an All-Layer Margin.

Heteroskedastic and Imbalanced Deep Learning with Adaptive Regularization.

GM-RKB WikiText Error Correction Task and Baselines

Federated Accelerated Stochastic Gradient Descent.

Exploratory longitudinal analysis of cfDNA to reveal potential biomarkers of CRC progression and treatment response

Evaluating Machine Accuracy on ImageNet

Essential Effects

Entity and Evidence Guided Relation Extraction for DocRED.

Document-Level Relation Extraction with Adaptive Thresholding and Localized Context Pooling.

Developing Metadata Categories as a Strategy to Mobilize Computable Biomedical Knowledge

Deep Learning Pocket Primer

Cinco meses de pandemia

Blood-based detection of early-stage colorectal cancer using multiomics and machine learning

Beyond Lazy Training for Over-parameterized Tensor Decomposition.